How we read your labs.

Flow

The tool runs four stages

Stage framework

Femme Fortified's 5-Stage Iron Deficiency Model

Several of the patterns in this tool map to a staged model of iron status, from Stage 0 (replete) through Stage 4 (iron deficiency anemia).

There is no formal diagnostic classification for iron deficiency from any leading health organization.

• WHO anchors only Stage 4 (anemia) to a specific hemoglobin cutoff. The pre-anemic stages are research terminology, not ICD-coded diagnoses. A clinician will not read "Stage 2" off your lab report.
• Iron status is a continuum, not a binary. There is no single agreed-upon cutoff that cleanly separates "enough" from "not enough."

The staged framing is what the peer-reviewed iron-biology literature uses. Papers on women's iron status over the last decade consistently describe iron status as a progression through stages. The classical three-stage clinical model (depletion → iron-deficient erythropoiesis → anemia) has been extended earlier by the identification of a compensation phase (our Stage 1), where the body begins upregulating intestinal iron absorption before stores are measurably depleted.

Where each transition comes from:

• Mei 2021 anchors the Stage 2 → 3 transition at ~25 μg/L ferritin in non-pregnant women.
• Galetti 2021 anchors the Stage 0 → 1 transition at ~50 μg/L.
• Mei 2025 extends the framework to pregnancy by trimester.

Every stage threshold in this tool traces to one of these papers. See the Evidence page.

We adopt this staged model because it is the framing the current research uses, and because it gives us a structure for naming pre-anemic iron states that standard reference ranges often miss. The plain-language labels on your results page ("Your iron levels look healthy," "Early signs of iron running low," and so on) are the user-facing versions of these stages.

Life-stage adjustments

Thresholds shift with physiology

Patterns

The 16 patterns this tool can recognize

Most results route to one pattern. When two or more patterns fit your values about equally well, we surface them as a mixed read — every applicable pattern named, the reasoning behind each shown, the combination left for you and your clinician to work through. Severity banner, narrative, question prompts, and retest recommendations are pattern-specific in either case.

Severity tiers

How we label clinical priority

Every pattern carries one of three severity tiers. The tier names the action, not the severity of your iron status; it tells you how soon the finding warrants a clinician's attention.

A "For your records" tier doesn't mean "no finding." It means the pattern doesn't require fast medical follow-up. You can work through the educational material on your own timeline and bring it up at a regular appointment if you want to.

How the tier is chosen. Severity is computed from the inputs and the classified pattern, not from a user-reported symptom score. The defaults:

• "See a clinician today" triggers are hard rules on lab values.
• "Bring to your next appointment" is the default for any pattern where the research suggests clinical interpretation adds value beyond education.
• "For your records" is the default for everything else.

Arbitration

When multiple patterns could apply

Rules fire in a deterministic priority order:

1. Overload flag — if very high ferritin + high TSAT, the overload pattern supersedes regardless of other findings.
2. Inflammation flag — if CRP ≥ 5 or recent illness reported, the inflammation pattern supersedes iron-deficiency patterns to avoid misreading acute-phase ferritin.
3. Anemia threshold — if Hb is below the life-stage-specific anemia threshold, the anemia pattern takes precedence over depletion patterns.
4. Depletion stages — among stages 0–3, the lowest stage the data supports is applied.
5. Fallback — if no iron-specific marker is provided, we report "not enough information to classify" and name the single test most likely to give a clear pattern read.

Evidence framework

How we badge the research

Rule derivation

Where each rule comes from

Every rule in this tool comes from one of three places. Each one is tagged in the audit trail so the source of any given output is reconstructible.

Direct paper thresholds. Most rules quote a specific number from a specific paper. When a rule has a single source, the paper is cited and the badge reflects that paper's evidence level. Examples:

• Stage 3 ferritin cutoff at 25 µg/L: Mei 2021.
• Trimester-specific anemia thresholds: WHO 2024 consensus plus Mei 2024.
• Overload evaluation triggers: EASL 2022.
• Hemoglobin severity cutoffs: WOMAN-2 trial analyses.

Multi-paper synthesis. Some rules apply a clinical-decision framework that spans more than one paper. Each rule cites every paper involved; the citation chain shows on the results page. Examples:

• Skikne 2011 differential framework (anemia plus non-deficient ferritin routes to an MCV-led workup) drives the anemia present, but iron stores look adequate pattern.
• Cappellini 2020 morphology review handles the microcytic-with-adequate-iron lane.

We tag rules as synthesis where a single paper does not anchor the full call.

Edge-case patterns. Three patterns name the limits of what the tool can say rather than a specific clinical state:

• Your results could fit more than one pattern fires when two corpus-supported readings genuinely apply at once.
• Not enough information to classify fires when the panel is incomplete — one or more of the four core markers (ferritin, hemoglobin, MCV, TSAT) is missing — and adding a marker would resolve the read. When this fires, the tool runs a forward-look on your specific inputs: it sweeps plausible values for each missing core marker, re-classifies against everything you already entered, and recommends the single marker most likely to land you on a substantive pattern. The recommendation is tailored to your inputs, not a generic prompt.
• Your combination doesn't match an established research pattern fires when all four core markers are present but the published research has no named pattern for the combination.

None of these invent a clinical claim. They flag where the tool itself reaches the edge of what the research has mapped, and route to a clinician with the lab values in hand.

Cross-validation. To double-check our rules, we built a second classifier from scratch. This second version follows only the published papers; it cannot see the tool's own code. We run about 64,500 test cases through both versions. On every case where the research gives a clear answer, the two classifiers reach the same conclusion. The independent classifier lives in our test suite, and we report the match rate each time we update the decision logic.

Limits

What the tool does not do

• Diagnose. Clinical diagnosis is outside the scope of an educational tool. We classify patterns. Clinicians diagnose.
• Recommend dose or form of treatment. We do not suggest oral vs. IV, ferrous vs. ferric, or any specific dose.
• Adjust for inflammation. BRINDA and Thurnham correction factors are population-level tools. We flag the confound and defer to clinical judgment rather than auto-adjust.
• Evaluate pediatric iron. Under-17 users are blocked at Step 1. Pediatric iron physiology differs materially.
• Screen for hemochromatosis. Patterns suggestive of overload route the user to clinical evaluation without further workup from this tool.

Versioning

The research evolves, and we do our best to keep up

This tool is versioned. Every PDF records the version that generated it, so a saved reading reflects the state of the research at the time it was generated, not the latest state.

We try to stay current with the published literature on women's iron status, and we update the logic when we find evidence that warrants it.

The current version is 1.0 (April 2026).

This methodology is a snapshot of the research at the time this tool's logic was generated. See the version date on any saved PDF to confirm which state of the research your reading reflects. This page describes how we read lab values; it is not medical advice.